The proposed study will utilize the longitudinal knee radiographs obtained in all OAI study participants to provide structural endpoints for users of the OAI public databases by: 1) comprehensively reading all of the knee radiographs for OA prevalence, progression and incidence using Kellgren and Lawrence (K-L) grades and individual radiographic feature scores; 2) obtaining quantitative measures of joint space width in all knees of subjects in the Progression cohort and knees with OA in subjects in the Incidence cohort. The results of these assessments will be released as public use datasets through the OAI Online public website. As noted in the RFP, the OAI cohort study is a publicly available database of baseline and annual follow-up assessments in 4,796 subjects who have, or are at risk of developing, knee OA. The OAI was designed for studies of imaging, biochemical and genetic biomarkers of OA incidence and progression, and studies of the natural history and risk factors for OA. The main focus of the OAI is on the knee. Standardized knee radiographs and MRIs of the knee were performed on all subjects at baseline and repeated in all subjects who attended the follow-up clinic visits and who remained eligible for knee imaging. In order to realize the full potential of the OAI, the proposed project will provide users of the OAI public database with a standard set of variables on the prevalence, incidence and progression of radiographic knee OA (RKOA) for use in biomarker (MRI-based imaging, biochemical and genetic), risk factor and natural history studies of knee OA. To perform this work, we will enlist experienced reading centers for knee OA structural outcomes with established track records in providing high quality data for large epidemiological and clinical studies. These teams have already been tapped by the OAI to provide central image assessment data on a limited scale in the OAI Progression cohort. The current project proposes to complete these central image assessments in the Progression subcohort and extend this work to the much larger Incidence subcohort.